. . . . . . "Open-source Sentinel-2 turbidity retrieval validated against in situ data in the Westerschelde estuary" . . . . . . "Five deliberate deviations, each recorded rather than hidden:\n\n1. Atmospheric-correction processor. The paper selected C2RCC for its turbidity\n result (the \"cN783\" chain, C2RCC + Nechad 2009). C2RCC exists only inside ESA\n SNAP and crashes natively in the container, so it is excluded; this replication\n runs Acolite instead, making the chain \"aN783\" (Acolite + Nechad 2009), an\n open-source analogue. Note the original study itself reports Acolite as its\n worst-performing atmospheric-correction processor, so a strong turbidity result\n obtained through Acolite here would be a notable outcome rather than an expected\n one.\n\n2. Site. Validation is in the Westerschelde estuary, not the Sado, because the Sado\n in situ match-up data are available only on request (paper Data Availability\n Statement) and no open, contemporaneous, in-estuary substitute exists. The\n Westerschelde is a comparable mesotidal, well-mixed, turbid estuary.\n\n3. Period. The record spans January 2016 to July 2026 (the full Sentinel-2 era),\n not the original's March 2018 to March 2020 field campaign, to gain a more\n complete set of match-ups.\n\n4. In situ reference. Coincident turbidity comes from Rijkswaterstaat routine\n monitoring stations, not the original AQUASado sampling campaign.\n\n5. Fully open-source chain. Every step (Sentinel-2 L1C, Acolite, the native Nechad\n 2009 turbidity product) is open and reproducible, whereas the original's\n selected chain depends on the proprietary SNAP/C2RCC processor." . . "The satellite half of the paper's turbidity pipeline is reproduced with a fully\nopen-source chain (the \"aN783\" chain) and validated against an independent in situ\nreference. Sentinel-2 A/B/C Level-1C granules over the Westerschelde estuary are\natmospherically corrected with Acolite to water-leaving reflectance, clipped to an\nestuary bounding box covering the along-axis stations. Turbidity is taken from\nAcolite's own native Nechad et al. (2009) turbidity product, the open-source\nanalogue of the paper's selected turbidity chain; the product is computed by\nAcolite with its published relative-spectral-response-convolved Nechad\ncoefficients rather than a hand-transcribed algorithm, and the near-infrared band\nnearest a 783 nm target is selected, matched by nearest wavelength to absorb the\nSentinel-2 A/B/C band-centre drift. At each in situ station a 3x3 pixel window is\ncut on the native 10 m UTM grid, with per-pixel quality control (water only, no\ncloud/shadow/glint, atmospheric-correction success), and the valid, finite pixels\nare averaged. Satellite retrievals are paired with coincident in situ turbidity\nfrom Rijkswaterstaat monitoring stations on station and time within a\nplus-or-minus two-hour window. Agreement is scored with the paper's metric suite\n(coefficient of determination, slope, intercept, RMSE, bias, unbiased RMS,\nabsolute and relative percentage difference); unlike the chlorophyll-a limb, the\nregression-family metrics for turbidity are computed on untransformed values in\nlinear space, not in log10 space, since turbidity does not span the orders of\nmagnitude that motivate the log transform for chlorophyll-a. Statistics are\nreported both over the full Sentinel-2 record and over the original study's\nfield-campaign window, to separate the effect of changing the period from the\neffect of changing the site." . "This study tests only the turbidity limb of the original claim: the assertion\nthat Sentinel-2 MSI retrieval of turbidity is the strong side of a parameter-\ndependent accuracy — that turbidity is highly consistent with in situ observations,\ndemonstrating the sensor's capability to monitor this water quality parameter. In\nscope is whether an independent, satellite-derived turbidity product agrees with\ncoincident in situ turbidity measurements in a mesotidal, well-mixed, turbid\nestuary, and whether the original study's strong-turbidity finding still holds when\nthe retrieval chain is fully open source and transferred to a different estuary and\na longer period. Out of scope, and left to separate atomic chains, are the\nchlorophyll-a, suspended-particulate-matter and coloured-dissolved-organic-matter\nlimbs of the paper's asymmetry, the seasonal time-series analysis, and any\ngrid-sensitivity (HEALPix / DGGS) representation of the corrected fields." . . . "Anne Fouilloux" . "2026-07-25T16:48:00.043Z"^^ . . . . . "Open-source Sentinel-2 turbidity retrieval validated against in situ data in the Westerschelde estuary" . . "RSA" . "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" . "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" . . .